EMERGENCY AMBULANCE. UTILIZATION IN THE CITY or LANSING, . MICHIGAN Thesis for the Dagree' of M. 'U. P. MICHIGAN STATE UNIVERSITY JAMES ALAN BREUKER 1977 .VI‘. .. I; ....k . u!» I.“ .19 I I ABSTRACT ESTIMATING EMERGENCY AMBULANCE UTILIZATION IN LANSING, MICHIGAN By James Alan Breuker This paper sheds some light on the problem of estimating the fre- quency of occurrences for ambulance runs in the city of Lansing, Michi- gan. Resources are being wasted and misused every day because of a lack of prOper planning in the emergency ambulance field, both in Lan- sing and throughout the country. Costs for providing emergency ambu- lance services are continually on the rise and a prudent and systema- tic way of planning for and utilizing emergency ambulances must be de- veloped. A relatively simple model, using census data and other data com- bined by census tract, is developed here. This model allows the user to project probable ambulance use by census tract. Thus, one can tell where emergency ambulances may be needed most and can plan for their locations accordingly. An example of this is the analysis at the end of this paper of the Lansing and Grand Rapids emergency ambulance sys- tems and how the research done in this paper might affect those sys- tems o By using census tract data in regression analysis, it was found James Alan Breuker that a four variable model, developed in this paper, could quite ac- curately project ambulance runs per person per year in the census tracts of Lansing, Michigan. These variables reflect the fact that, in most cases, ambulance runs have tracts with high percentages of di- vorced men as their destination. They also have as their destination tracts with high percentages of people of lower educational levels, of lower income and with higher numbers of auto accidents. Of course, all four of these characteristics do not have to be present in every tract with a high ambulance run rate, but usually at least one or a combination of them is. As a result, the four variable model developed has taken on the following form (numerical values have been rounded). Ambulance runs per person by tract - 1.48 (divorced men per persOn in each tract) + (-.Ol)(actual median school years completed in each tract) + .000003 (actual median income for each.tract) + .00009 (actual auto accidents for each tract). All the variables in the equation can be considered measurements of socio-economic class or social disorganization. EMERGENCY AMBULANCE UTILIZATION IN THE CITY OF LANSING, MICHIGAN By James Alan Breuker A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF URBAN PLANNING Department of Urban Planning and Landscape Architecture l977 This Thesis is dedicated to my wife, Sharon, without whose help and un- failing support this document could not have been completed. 11 goes ACKNOWLEDGEMENTS A special note of thanks for their help in preparing this thesis to: Dr. Roger Hamlin, my patient advisor, Professor Donn Anderson, whose constructive criticism helped to make this a better paper, Mr. John Kessler, for his continual support, and Mr. Paul Stuhmer, my fellow student, who helped me understand what I did not under- stand before. iii TABLE OF CONTENTS LIST OF TABLES ..................................................vi LIST OF FIGURES ................................................vii INTRODUCTION .....................................................l Chapter Introduction .............................................6 General Historical and Legal Background ..................6 Ambulance Service in Lansing .............................9 Historical Background ....................................9 m.t OOOOOOOCOOOCCCOOOOOOOOOOOOOOOOOOOOOOCOOO0.0.0.00000014 $~ry 000......OOOCOOOOOOOOOOCOOOOOOO00.00.000.0000000016 11. REVIEW OF LITERATURE AND INFORMATION SOURCES ..............17 Introduction .........,..................................17 Local Sources of Information ............................17 Literature and Library Search ...........................18 Summary .................................................21 111. mommY OOOOOOICCOOOOOOOOOOOOOOO.00...00.0.00000000000024 Introduction ............................................24 Study Development .......................................24 variables ...............................................26 Census Tracts Used ......................................14 Summary .................................................36 IV. nsm’Ts 0.000....OOOOOOOOOOOOOOOOOOO00....00.00.00.00000000‘8 Introduction ............................................18 Computer Method .........................................18 Correlations ............................................39 The Equation ............................................05 Discussion of Results ...................................5? General Problems and Assumptions ........................55 Analysis of Results .....................................59 Application of the Model ............... ...... ...........64 Summary ...................... ..... ...... ....... .. ..... ..69 iv COMUSION OOOOOOOOOOOOOOOOOOOOOOOO00.......0.00.0.000000000000000071 APPENDIX .COOOOOOOOOOOO0.0000000000000000... O0.00.00.00.0000000000073 BIBLImmHY .OOOOOOOOOOOOOOOOOOO0.00.... 0.... 000000000 0.0.0.0....86 General References ........................ 00.0.000000000000000088 Table LIST OF TABLES variables Used in Collecting Data ........................... Multiple Regression Summary Table ........................... Correlation Coefficients .................................... Summary Table I of Model variables .......................... Summary Table II of Model Variables ......................... 0bservation--Census Tract ................................... Actual-Estimated Ambulance Runs, 1975 ....................... Actual-Estimated Runs Per Person, 1975 ...................... Estimated Emergency Medical Calls for Grand Rapids, Michigan. V1 28 (.0 62 46 49 52 76 78 84 LIST OF FIGURES Figure I. 2. 3. a. 5. 10. B’le Lansing Fire Department Ambulance Locations - 1975 ...........12 Lansing Fire Department Ambulance Run Increases, 1965-1975 ...11 Increases in Population and Ambulance Runs, 1970-1975 ........15 Lansing Census Tracts Not Included in Study ..................36 Y Estimate Calculations and Comparison of Estimates to Actual 0..O...OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.00.51 Lansing Census Tracts With 50 or More Ambulance Calls Per 1“ P.r.on.’ 1975 OOOOOOOOOOOOOOOOO0.00...000.000.000.000000057 Lansing Census Tracts With 50 or More Ambulance Calls Per 1000 Persons, Locations of Firestations, and Locations of Firestations Housing Ambulances - 1975 .......................60 Lansing Census Tracts With 25 or More Ambulance Calls Per 1000 Persons, Locations of Firestations, and Locations of Firestations Housing Ambulances - 1975 ....................62 Grand Rapids Census Tracts with 25 or More Projected Emer- gency Medical Calls Per 1000 Persons Per Year ................67 Grand Rapids Census Tracts with 50 or More Projected Emer- gency Medical Calls Per 1000 Persons Per Year ................68 Actual Versus Estimated Ambulance Calls Per 1000 Persons By Census Tract, Lansing, Michigan, 1975 .....................79 vii INTRODUCTION In recent years, there has been much interest in the field of emergency medicine, especially emergency ambulances. Many studies have noted the effectiveness of advanced equipment and well trained person- nel in saving lives and preventing injury.l While no one can argue that emergency ambulances have not been beneficial, the question of whe- ther or not they can be more effective must be explored. To cities which operate their own ambulance system, increasing the effectiveness of ambulance service is important if ambulances are to aid the most people for the least cost. Unfortunately, the methods which many cities use to estimate emer- gency ambulance utilization are not well formulated. Many cities sim- ply use past trends in ambulance utilisation to project future use. While this method is better than no method at all, there is much more to emergency ambulance utilisation than just what past trends indicate. Using more scientific methods can be of great importance in helping emergency medical services planners avoid hit and miss solutions to emergency ambulance problems. This paper is an attempt to help answer the question of where 1 "Hospitals on Wheels Replaces Undertaker-Sponsored Ambulances,“ American City, vol. 88, May 1973. p. 119. "Paramedics Give Fast, Pro- fessional Emergency Care," American Cit , vol. 89, March 1976, pp. 55- 56. ambulances should be placed to maximise their effectiveness. The way this paper answers this question is by proposing a computerised model for projecting ambulance utilisation by multiple linear regression an- alysis. The model was then applied to the city of Lansing, using data from Lansing Fire Department, Lansing Police Department, Lansing Traf- fic Bureau and 1970 Census of Population to gather information by cen- sus tract. The entire emergency medical services system,.i.e., emergency rooms, emergency clinics, private ambulance companies, etc., is not considered here; only the ambulance service operated by the Lansing Fire Department. Because of this, data only from lensing Fire Depart- ment was used in analysing ambulance utilisation. It must be kept in mind that this study is concerned with pro- jecting ambulance runs per person by census tract and, as such, no statement is made or intended on other related issues such as train- ing, equipment, or quality of care. Something will be said concern- ing location analysis at a later point in the paper. This study can suggest some answers concerning why ambulances go where they do most often. It can be valuable in helping a municipal- ity analyse its current system. Also, if a city or private ambulance service intends to begin operations in a place which has not had ade- quate emergency ambulance coverage before, this study can be of aid in deciding where_to place ambulances so they can serve the most people for the least cost. This paper consists of six main parts: Background, Review of Lie- eraturs and Information Sources, Methodology, Results, Conclusion, and the Appendices. The Background section explains some of the historical background of emergency ambulances and how they evolved into what they are today, including some of the legal requirements for operation. Specific his- tory concerning the Lansing ambulance service is also given. An ex- planation of how the system currently operates is given in this sec- tion, too. This part of the paper is meant to help the reader better understand how an emergencyyambulance system operates. Thus, the re- mainder of the paper can take on additional meaning to those who may not be extremely familiar with the workings of an emergency ambulance service. The section dealing with review of literature summarises which sources of information were considered in the research for this paper. People from local agencies and organisations were contacted and inter- viewed and letters were sent to others seeking information on the to- pic of emergency ambulance utilisation by census tract. A library search was also instituted which covered periodicals as well as books. The Council of Planning Librarians bibliographical publications were checked and the Tri-County Council for Emergency Medical Services did a computer bibliography search. Their search included what the Depart- ment of Health, Education and Welfare had in its computer bibliography on the topic of emergency ambulance services utilisation. The results of the entire search for information are included in the Review of Lit- erature and Information Sources part of the paper. The Methodology section deals with the process whereby this study was completed. An explanation of how this study was devised as well as how it was carried out is given and a detailed account of which variables were used in the study is also given. A short-outline of the computer process necessary for this study and its effect on the variables and the number of cases (census tracts) used in the study is presented, too. In addition, the process of aggregating varia- bles not found in census data is explained. The next section of the paper, Results, is concerned with pre- senting, explaining, and discussing the results obtained by this study. Included in this section is both an explanation of the com- puter method used and an explanation of the model used to project am- bulance runs per person by census tract. After this explanation comes a discussion of the results. Following this discussion, there is a section on some of the problems which this study encountered. The final section analyses the impact of applying this model to Lan- sing and Grand Rapids. The Conclusion summarises what has been accomplished in this paper and emphasises the need for more research. Conclusions about the methodology and results are drawn and some implications of this study for future research are stated. Appendices are included in this paper to help explain the accom- plishments of this study._ Several of the appendices were compiled by using figures derived from the model and they show how the model can be applied. This paper was researched during the school years 1974-75 and 1975-76. Much information was gathered from the Lansing Fire Depart- ment, Lansing Mercy Ambulance Service, Tri-County Council for Emergency Medical Services, The Office of Health and Medical Affairs--State of Michigan, Michigan Mid-South Health Systems Agency, Lansing Police Department, and Lansing Traffic Bureau. Background papers were re- searched and completed during the fall of 1975. Data collection was completed during the spring and summer terms of the 1975-76 school year and the calculation and computation process was done during the summer of 1976. The writing was done in the summer and fall of 1976. CHAPTER I BACKGROUND Introduction Since this study concerns ambulances, and uses the City of Lan- sing's ambulance service as a case study, it may be helpful to know more about the general history of ambulances as well as the more speci- fic history of the Lansing service. It can also be helpful to know how the current ambulance service in Lansing operates and how effective it has been. Regulation of ambulance services by federal and state governments is discussed here as is the question of instituting more complete planning methods for an ambulance service. General Historical and nggl Background Ambulances have been in existence for quite some time, although the form of today's ambulance system and equipment has changed from the original. Most ambulances were originally associated with wars and were concerned with transporting wounded soldiers to field hospitals. Gradually, the idea of using ambulances spread to the civilian world, also. Traditionally, the only person to have a vehicle large enough to transport sick or injured comfortably was the mortician or the underta- 2 ker. Many morticians began to operate an ambulance service for the 2 Mike Amble, interview held in office of Tri-County Council for Emergency Medical Services, September 1975. 6 benefit of those families which had previously used the mortician's services. The ambulance service quickly spread to include the whole community and funeral homes began to charge for their services. At first, the method of transportation of sick and injured was quite sim- ple. An ambulance would simply go to where the person in need of ser- vice was, pick him up, and transport him to a hospital, hoping that he would not die on the way.3 Any treatment administered to the patient at the scene or enroute to the hospital was usually given by ambulance attendants who had no training or were trained only in Red Cross first aid.4 The only regulation of the "ambulance industry" pertaining to attendants' qualifications was the requirement that ambulance attend- ants be trained in advanced Red Cross first aid. This requirement was not enacted into law until 1969.5 Prior to that time, any train- ing for ambulance attendants was mostly voluntary on the attendant's or their employer's part. The Red Cross advanced first aid requirement still stands, although Michigan Senate Bill 986, passed by the Senate in March of 1976, would require all ambulance attendants to be trained to the basic emergency medical technician, or EMT-1, level. This legis- lation is expected to be implemented in the near future. Michigan Public Act 258 of 1968 requires updating of ambulances and equipment. The Michigan Department of Public Health has set up guidelines for ambulance services to follow when purchasing or updating 3 Interview held with a former ambulance attendant, Estes-Leadly Funeral Home, Lansing, Michigan, November 1974. l. Ibid. 5 Michigan Public Act 260 of 1969. 6 their equipment. The federal government has no legislation pertain- ing to training and equipment for those wishing to use 0.8. Government money to purchase new ambulances. The National Academy of Sciences has set up guidelines for training ambulance personnel, but these are not legally required.7 New requirements, at least in the State of Michigan, are likely to be enacted in the near future.8 These requirements mean a greater chance of survival for the medical emergency patient, but they also mean increased cost to the consumer and to those providing the service. In addition to the prospect of increased future regulation and, there- fore, increased cost, is the prospect of competition between ambulance services. Ambulance services can advertise and are responsive to pub- lic demand for increased quality of care. This, in turn, also means greater cost. As a result, many funeral home services have dropped out of the business since the ambulance service was not their primary busi- ness. Some still operate in rural areas where the competition is not as keen, but many in urban areas are phasing out their operations or 9 are getting into it more as a business rather than a sideline. As a 6 Telephone interview held with Lori Martindale, secretary for the Tri-County Council for Emergency Medical Services, August 1976. 7 National Academy of Sciences--National Research Council, Division of Medical Services. Training 2;,Ambulance Personnel egg Others Respgn- sible §2£_Emergengy Care of the Sick and Injured 25 Egg Scene 522 22;- lgg Transpgrt, U.S. Government Printing Office, 1968. 8 Michigan Senate Bills 984, 195, 986, and 987, passed by the Michi- gan Senate in March of 1976. 9 Interview held with a former ambulance attendant, Estes-Leadly Funeral Home, Lansing, Michigan, November 1975. result of this, some cities are being forced to provide ambulance service themselves or contract for it from commercial services. Ambulance Service 12_Lansing: Historical Background The City of Lansing is required by its current charter to provide ambulance service to its citizens. The charter does not stipulate that the city must run the service. The city must only make sure it is pro- vided. The service could be contracted for, but the city has chosen to provide the service itself. Lansing's ambulance service was first operated by the police de- partment. In the early 1950's, many patrol beats for policemen were foot patrol beats. Therefore, ifra call came for the ambulance, the dispatcher would have to locate a foot patrolman and he would then have to run downtown to the police station, get in the ambulance, and proceed to the aid of whoever was in need. This process took a long time, but at that time the city was much smaller than it is today. In addition, the service provided was simply one of transportation to a hospital or simple resuscitation with an inhalator or oxygen equipment. As the city grew, the demand for ambulance service did also. Coupled with this, was the demand for upgraded service. Since police are deemed more effective when they are patrolling, the police depart- ment did not tie up manpower at the station by permanently assigning men to the ambulance.10 Therefore, the ambulance service was shifted to the fire department, which had manpower more readily available on a twenty-four hour basis. 10 Elmer Neisenber, dispatcher for Lansing Fire Department, inter- viewed November 1975. 10 At first, the fire department used the same vehicle and equipment as the police department did and provided the Same service also. As demand changed and as new breakthroughs in emergency medicine were brought about, the fire department made every effort to keep in step. Since the department requested federal money for vehicles, it had to meet federal standards and this meant an improvement in equipment. Men also were required to be trained to use the equipment and, because of this, the quality of the ambulance service in Lansing seems to have im- proved. The service began with one vehicle and is now operating with four. Three are always in service and one is as a backup. The city has a new "mobile emergency room“ on order which is larger than a regular ambu- lance and is better equipped to handle a greater range of medical emer- gencies. Lansing is also in the process of implementing paramedics on all its ambulances.ll The paramedic program enjoys a broad base of pub- lic support. Many of the regularfiremen have been trained in advanced first aid and the city has plans to train as many as it can to the higher level of Emergency Medical Technician I (a paramedic is an Emer- gency Medical Technician II).12 The city's ambulances are stationed at different fire stations throughout the city. The back-up ambulance is stationed at the down- town station since there is more room there and, since that station's ambulance receives more calls than the others, that is where an 11 Interview with Pete Decker, Assistant Fire Chief, Lansing Fire Department, November 1975. 12 Ibid. 11 additional ambulance is likely to be used most effectively. The down- town station is also central to the city; therefore, calls from outly- ing areas can be handled easier and more quickly than if it were sta- tioned elsewhere (see Figure 1 for ambulance locations). Since the city's ambulances are supposed to respond only in cases of emergencies, the dispatcher must determine if it is a true emergency. If he determines that it is an emergency, he must send out the ambulance nearest to the incident. If he thinks the call is not a true medical emergency, he refers the call to a private ambulance service which re- sponds to it. Ambulances which respond to emergencies also transport the victims to a hospital, unless specifically requested not to do so. Finances for operating the service come from city revenues as well as federal and state grants for training and equipment. The equipment and training are kept as current as possible. The fire department closely adheres to federal and state guidelines and recommendations con- cerning training and equipment. Planning for new vehicles or extended service is not an elaborate process in the city of Lansing, nor is it really necessary yet. Deci- sions for additional vehicles or manpower seem to be guided by past trends in demand and not according to future projections. Each year the fire department has experienced an increase in the demand for ambulance service (see Figure 2, page 11). The fact that the user pays no direct fee for service (he/she pays through taxes), that the ambulance service enjoys much public support, that the availability of ambulance service is becoming more well known, and that paramedics serve rather than less trained attendants all add to the increasing demand for the service. 12 3101 I __J 3 2 3‘ 11.02 ‘ Ambu lances located a t f irestat ions % 2M Figure 1 LANSING FIRE DEPARTMENT AMBULANCE LOCATIONS -- 1975 15 AMBULANCE RUNS 2000 1000 1965 1966 1967 1966 1969 1970 1971 1972 1973 1975 1975 YEAR Figure 2 LANSING FIRE DEPARB’EENT AMBULANCE RUN INCREASES, 1965-1975 SOURCE: lensing Fire Department 14 Population in Lansing is estimated to have increased from 131,114 in 1970 to 131,545 in 1975, or about 21,13 while ambulance calls increased from 3.169 in 1970 to 4,671 in 1975 or about 321 (see Figures 2 and 3. pages 13 and 15) 19 This means an increased workload and the city feels it may have to add an additional ambulance to meet increased demand. The question of the location for the additional ambulance is as yet unanswered. In order for the city to plan more accurately for future utilisa- tion of its ambulance system, it should use more comprehensive and scientific methods. Using other data besides past trends can be of great importance in helping emergency medical services planners avoid hit and miss solutions to emergency ambulance problems. Relying on past trends of continual increase in demand may not be an adequate in- dicator of future use. Egg; Like many other services today, the cost of ambulance service is increasing and it uses resources that possibly could be utilised better in another area if proper foresighb is used in planning. However, the ambulance industry has not been stabilised by implementation of uniform 15 standards and, as a result, costs have not leveled off. Lansing 13 The 1970 figure was obtained from the Census while the 1975 fig- ure is a projection estimated by the Tri-County Regional Planning Com- mission. It was obtained during a telephone interview in June 1976. 14 These figures are from a telephone inquiry to L.F.D. in February 1977. 15 Telephone interview with Rod Palmer, owner of Lansing Mercy Ambulance Service, October 1976. 15 E058 «HE 6533 as 83358 gm inflame ESTES 33$»? .mznm mg 95. SEER 2H BEBE m egg 2.: on: . as: One— c o coc— 95.8— 95... 98.02 89. 98.9.. aeassuoi sauce—ac..— 95¢ a a u x 11.2 . ace» ”H258 l6 Mercy Ambulance Service, for example, experienced about a 332 increase in cost from 1975-1976 or an increase in their base rate from $15 to $50.16 In view of this, utilising resources to the optimum with fore- sight is imperative. More complete planning for emergency ambulance runs is prOposed in this paper. Thus, by using this method both to analyse present needs as well as to get a grip on future needs, the resources in an emergency ambulance system can be put to good use. Summary Ambulance service in Lansing has come a long way. It now operates with a much higher level of efficiency and expertise than it did in the past. The ambulance service enjoys a high degree of public support and this also has increased its effectiveness. However, the cost of pro- viding service and the increased demand for that service necessitates a more comprehensive means of planning for emergency ambulances and their utilisation. 16 Ibid. CHAPTER 11 REVIEW OF LITERATURE AND INFORMATION SOURCES Introduction This chapter explains what sources were consulted in gathering in- formation for this thesis. The search for information included library sources such as books and periodicals, local community sources such as agencies, commissions and councils serving Lansing and the surrounding area, and agencies and offices operated by the State of Michigan. Let- ters seeking information were also sent to people not in the immediate area who might be helpful in gathering information for this paper. £2521 Sources 2; Inforggtion This study was conceived after several months of working with and doing research on the emergency ambulance system in the tri-county (Clinton, Eaton, and Ingham Counties, Michigan) area. In working and talking with people at the Lansing and East Lansing Fire Departments, Michigan Mid-South Health Systems Agency, Tri-County Council for Emer- gency Medical Services, Office of Health and Medical Affairs for the State of Michigan, and the owner of one of the commercial ambulance services in the city, it became apparent that a study of the nature of this one would be beneficial. Therefore, after consulting with Dr. Roger Hamlin of the School of Urban Planning, Mr. John Kessler, Direc- tor of Planning at Michigan Mid-South Health Systems Agency, and 17 18 16 others, it was decided that a study of emergency ambulances and how their utilisation might be projected using readily available census da- te would be beneficial to the parties involved and the field of emer- gency medical services planning. Literature and Library Search To learn as much about studies of this type as possible and to be- come further familiarised with the subject area, a search of literature was instituted. The search for similar studies included a library search of the Eduggtion Index from July 1965 to May 1976. No litera- ture was found which would have a direct bearing on this paper. The topics which were searched in the Eduggtion Index and in all the follow- ing potential sources of information were: "Medical Services," "Emer- gency Squad Personnel," "Ambulances," "Rescue Squad," and "Medical Workers." Much the same topics were investigated in the Rggder's Guide. pics which might relate to this study, also. volumes from 1965 to May of 1976 were covered. Articles on the ambulance systems of Houston, Texas, Los Angeles, California, and Jacksonville, Florida were found, but none mentioned how these systems were planned or utilised.l7 Research (Resources) i3 Education sourcebooks were also examined. The period 1968 through May of 1976 was covered. This search produced 16 Dr. Harry Perlstadt, Department of Sociology, M.S.U., Mr. Mike Amble, Executive Secretary, Tri-County Council for Emergency Medical Services, and Mr. Kenneth Malkowski, formerly a health planner and now a transportation planner for the State of Michigan. 17 "Hospitals on Wheels," American City 88(May l973):1l9. "Parame- dics Give Fast...,” American City 89(March 1974): 55-56. “The Best Ambulance Service," Reader's Digest 104(March 1974): 95-98. 19 a selected bibliography, published by the U.S. Government, on emergen- cy health services.18 Unfortunately, this bibliography dealt with the general topic only and offered no clues for additional reading and re- search on the tapic of emergency ambulance utilisation and distribution. Th5 Egrrent Index 52 Journals 19 Education was searched. The period from 1969 to May of 1976 was covered and no similar studies were found here either. Council of Planning Librarians planning bibliographies which per- tained to emergency medical or health services were also researched. Those searched were: #100--P1anning For Locational Change Lg Egg Deliv- ggy 9; Medical 9555, #134--Hea1th Manpower Planning, #168--Hea1th Plan- BIBS: s233--§551£h_glgnning Applicgtions g; Qpergtions Resegrch 59g Systems Anal sis, #26l--Physicians Assistants, #392--Comgrehensive Hgglgh Plannin , #475--P1anning Medical £555, #493--Emergency Medical Services 15 Metropglitan 55535, #586--§gglth Services, #593--Ihg £32152 Compgnent £2 Community Development, and #690--Police and Fire. This search produced nothing which would relate to emergency ambulance utili- sation by census tract. Some general background sources were found and these have been included in the "General References" section of the Bibliography. Mr. Kenneth Malkowski, Health Planner for the Office of Health and Medical Affairs, State of Michigan, was consulted concerning possible 18 National Academy of Sciences--Nationa1 Research Council, Division of Medical Sciences. Recent Publications 22 Ambulgnce Services, Washington, D. C.: NRC-HAS, June 1970. 20 sources of information.19 He supplied material from his personal files20 (these are included in the bibliography) and supplied a book which was very useful in this study. This was a draft copy of the book entitled "Methods for Determining and Projecting the Needs and De- mands for Emergency Medical Services: An Anotated Bibliography," pre- pared by Arthur Young and Company for the Division of Comprehensive Health Planning, Health Resources Administration, Department of Health, Education and Welfare.21 This book proved its usefulness by providing an up to date list of research in emergency medical services. In addi- tion, this book is an anotated bibliography and a short summary of each article listed was given. This made it much easier to judge which po- tential sources of information were the most valuable. The Fitssimmons and the Aidrich, et al., studies, cited and discussed on pages 22 and 25 were found in this book. Some general background references also were found and these have been included in the "General References" section of the Bibliography. The Tri-County Council for Emergency Medical Services also did a 19 Interview held at the Office of Health and Medical Affairs, State of Michigan, Lansing, Michigan, June 1976. 20 Thomas P. Dekar, "A Simulation Model of Response Characteristics of the Detroit Emergency Medical Service" (Ph.D dissertation, Universi- ty of Detroit, 1976). Idem, "Proposal for Dissertation Involvement with Emergency Medical Service Division of Detroit Fire Department“ (Proposal for Ph.D dissertation, University of Detroit, September 1973). State Health Planning Advisory Council Task Force on Health Personnel and the Office of Health and Medical Affairs, Bureau of the Budget, State of Michigan, "Pharmacy Personnel Requirements in Michi- gan; A Preliminary Report,” draft copy, April 1976. 21 Arthur Young and Company, "Methods for Determining and Project- ing the Needs and Demands for Emergency Medical Services: An Anotated Bibliography," draft capy, 1975. 21 computer check of the sources it has available in its computer biblio- graphy. Its computer bibliography is connected with the computer bi- bliography on emergency medical services at the Department of Health, Education and Welfare in Washington, D.C. Nothing was found there about projecting emergency ambulance use by census tract data.22 In addition to the above, a letter was sent to several people who might have some additional knowledge in the field and who might be will- ing to share it or suggest further sources of information. A copy of this letter is shown in Appendix A. The names of these people were gathered from interviews of the people previously mentioned, references in periodicals consulted, and from replies received from others to whom the letter was sent. Sources encountered through this mail survey did not directly re- late to projecting emergency ambulance run rates by census tract using socio-economic data. Some of the suggested articles dealt with relat- ing emergency room visits to socio-eoonomic data,23 but not ambulance use. Therefore, these articles were consulted and their relationship to ambulance utilization was considered. Geoffry Gibson has written an interesting book entitled Emergency Medical Services in the Chicago Area in which he discusses the state of 22 Interview held with Mr. Mike Amble, Executive Secretary, Tri- County Council for Emergency Medical Services, June 1976. 23 Jacobs, A.R., J.H. Garett, and R. Harsinger, "Emergency Depart- ment Utilization in an Urban Community: Implications for Ambulatory Care." Journal 2; the American Medical Association, Vol. 216, no. 2, April 2, 1971, pp. 307-312. Berman, J.I. and Luck, E., "Emergency Services: Patients Ethnic Backgrounds Affect Utilisation," Hospitals, vol. 45, July 16, 1971, pp. 64-8. 22 24 the Chicago E.M.S. system. While he goes into how each vehicle is utilized and how well the equipment is used, no attempt is made to de- termine if the utilization rates can be estimated or projected or how this might be done. There are other studies of this nature also which, unfortunately, do not discuss using data outside the ambulance system itself to estimate calls. Rather, these studies analyze data gathered from inside the system, such as response time, to point the way toward increased efficiency.25 Studies of these types are noted here because they were reviewed in order to give the author some idea of what they did, how they did it, and whether or not they have a bearing on studies similar to the one presented here. The general format of these studies (hypothesis, research, collecting data, testing, and results) is rough- ly the same as the format of the thesis presented here. On the other hand, their intent is different from the character of this study. They look at ambulance utilization from the standpoint of how each vehicle and its personnel can be used better in the location it is in. The cur- rent location of the vehicles is assumed as appropriate. These studies attempt to focus on utilization efficiency using factors such as time spent on runs, equipment used on runs and number of personnel needed. This study proposes a method of evaluating ambulance location and uti~ lization efficiency by collecting census tract information which 24 Geoffry Gibson, Emergency Medical Services in Egg Chicago Area (Chicago, Illinois: Center for Health Administration Studies, Univer- sity of Chicago, 1970). 25 J.A. Fitzsimmons, "A Methodology for Emergency Ambulance Deploy- ment," Management Science, vol. 19, no. 6, February 1973, pp. 627-616. 23 relates to socio-economic data and using that data to project ambulance runs per person in census tracts. Thus, the location of ambulances can be planned so that those who use and need an ambulance most can be served the fastest. Summer Many possible sources of information concerning emergency ambulance utilization were consulted. Very little was found about projecting am- bulance utilization through the use-of census tract data. Other studies were found which used factors such as response time, but the use of socio-economic factors in projecting ambulance run rates has not been incorporated into the methodology for any study except the Aldrich, et al., one in which emergency ambulance use in Los Angeles is scrutinised. This study is discussed at greater length in the last chapter of this paper- CHAPTER III METHODOLOGY Introduction The topic for this paper was worked out with community leaders in the field of emergency ambulance services while the methodology for do- ing this study evolved from several months of work in the area of emer- gency ambulance services, from a related study done in Los Angeles, California, and from assistance given by Dr. Roger Hamlin, School of Ur- ban Planning and Landscape Architecture, Michigan State University, and Mr. raul Stuhmer, a fellow student. Study Developggnt Deve10ping a technique to project ambulance runs per person by any means.can be of great value to a city or ambulance service in planning for the future. Therefore, since there is much information concerning the population of Lansing gathered by census tract in the decennial U.S. Census, it was decided that attempting to project ambulance usage by utilizing this data would be a beneficial achievement which could be adequately accomplished. This decision was arrived at with Mr. John Kessler from Michigan -Mid-South Health Systems Agency and Mr. Ken Malkowski, health planner for the State of Michigan. Mr. Malkowski and Mr. Kessler suggested that something could be done on ambulance utilisation since this was an 24 2S understudied subject, especially in the Lansing area. The author then decided to do a preliminary study of ambulance utilization in the Lan- sing area. In the process of doing this, it was discovered that the Lansing Fire Department had begun to keep records on ambulance runs by census tract. With this in mind and knowing that there was much infor- mation in the Census, a brief study was undertaken to see if there might be a relationship between tracts with high ambulance calls and tracts with certain types of socio-economic characteristics such as low median school years completed, low median income, and a high rate of persons over 65 years of age. Simple visual observations indicated that there did seem to be some sort of relationship between ambulance runs and certain census information. It was then decided that the technique would be expanded and a model developed which might help one project ambulance runs using readily available census tract data. Having decided to use census tract data, the next step was to choose which census tract data to use. Some suggestions as to who uses an ambulance most, which neighborhoods the ambulance goes to most often and who uses other emergency medical services most often were gathered 26 27 from talks with L.F.D. paramedics as well as from printed studies. 26 L.F.D. paramedics Bill Wilson and Roy Zigler and Lansing Mercy Ambulance Service employees were asked where ambulances seem to go most often and who they most often picked up. Their answers indicated that it seemed that lost runs went to lower socio-economic class neighbor- hoods and also there seemed to be a lot of runs to pick up people of older ages, most of them women. 27 Carole Aldrich, John Hisserich, and Lester Lave, in their study entitled "An Analysis of the Demand for Emergency Ambulance Service in an Urban Area," American Journal 2; Public Health, vol. 61, no. 6, June 1971, pp. 1156-1169, note that children and elderly, single men (includ- ing unmarried, separated, and divorced), and non-whites generate more calls than their counterparts. Barry King and Ellis Sox in "An 26 Whenever these suggestions coincided with available census data or da- te that might be easily aggregated into census tract format, it was in- cluded as a variable in this study. Usually census data could be found which coincided with these suggestions or closely paralleled them.28 Other variables were chosen because they seemed to suggest that they may have some bearing on ambulance use, e.g., the number of housing units per tract with no car available may indicate why so many people call an ambulance when one is needed rather than drive to the hospital. Variables The method used to analyze the variables is multiple linear re- gression. This method allows the user to establish the variables in a list from the most important to the least important. Inter-relation- ships between variables are considered and, when the computer is used, one can calculate the chance that a correct answer will result if a cer- tain piece of data is used in an equation. The multiple linear regres- sion technique develops a set of values which can be placed in a multi- ple linear regression equation. These values in the equation can then be used to develop a model which projects, in this case, emergency am- bulance use per person by census tract. A more complete explanation of this computer method is given in the first section of Chapter Four. Emergency Medical Services System-~An Analysis of Workload," Public Health Reports, vol. 82, no. ll, November 1967, pp. 995-1008, and A.R. Jacobs, J.W. Garett, and R. Wersinger in "Emergency Department Utiliza- tion in an Urban Community: Implications for Ambulatory Care," Journal 2; the Americgg Medical Association, vol. 216, no. 2, April 2, 1971, pp. 307-312, also indicate that various socio-economic factors influ- ence use of emergency medical facilities. 28 If census data did not directly coincide with a desired variable, then two or more census categories were combined which gave the desired 27 Table I indicates which variables were used in this study. The dependent variable is ambulance runs and is explained a little later in the text. The independent variables are listed following the depend- ent one. Most of the variables have been changed so they represent a rate or per person description. This was done to standardize the variables in each census tract as much as possible, i.e., the fact that a certain tract may have a high number of divorced males may lead one to form a certain opinion about the tract. When that factor is changed to a rate, it may be less visible than in other tracts. Rates automa- tically account for the amount of population in a tract and its influ- ence on certain variables or characteristics. Actual median income in each tract and actual median school years completed in each tract are used, since these are already in "per person" form. As they are, they indicate the relative poverty or wealth of a census tract. The unem- ployment variable is also a rate, but is calculated with the total num- ber of peeple in the work force over 16 years of age as the denominator and not all peOple in the tract as the denominator because not all peo- ple in a tract necessarily want to be in the work force. The housing variable was intended to give some idea of the number of crowded and/ or below standard houses in each tract. This variable was calculated by adding dwelling units with 1.01 persons per room or more in each tract to the number of dwelling units lacking some or all plumbing fa- cilities in each tract and dividing by twice the number of households. Another variable which is not calculated on a per capita basis is the "One Adult Headed Family“ variable. This variable was computed by variable, e.g., women over 65 + men over 65 a persons over 65. 28 Table 1 VARIABLES USED IN COLLECTING DATA variable Name Dependent variable: Ambulance Runs Independent variables: Females Persons Under 10 Years Old Females Over 65 Years Old Males Over 65 Years Old Median Income Police Calls for Service Non Whites Median School Years Completed Variable Code Name Runs Female Kids Oldfem Oldmen Income Cops Manor Skool Description Number of Ambulance Runs by Lansing Fire Department per Capita Per Year in Each Cen- sus Tract Number of Females Per Capita in Each Census Tract Number of Persons Under 10 Years Old Per Capita in Each Census Tract Number of Females Over 65 Years Old Per Capita in Each Census Tract Number of Males Over 65 Years Old Per Capita in Each Census Tract Actual Median Income In Each Census Tract Number of Police Calls for Service Per Capita in Each Census Tract Number of Non-Whites Per Capita in Each Census Tract Actual Median School Years Completed Per variable Name Independent variables: Unemployment Auto Accidents Divorced Females Divorced Males Housing One Adult Headed Families No Available Auto Poverty Level 29 Table l -- Continued variable Code Name Unem Crash Divfem Divmen Hous Singfam Nocar Poverty Description Census Tract Number of Unemployed Males and Females in the Labor Force Over 16 Years Old Per Capita in Each Census Tract Number of Actual Auto Accidents in Each Census Tract Number of Divorced Fe- males Over 14 Years Old Per Capita in Each Cen- sus Tract Number of Divorced Males Over 14 Years Old Per Capita in Each Census Tract 1.01 Persons Per Room or More in a Dwelling Unit Plus the Number of Dwelling Units Lacking Some or All Plumbing Facilities Per Twice the Number of Households in Each Census Tract Numbers of Families With Only One Adult As Head Per Total Number of Households with Heads in Each Census Tract Numbers of Housing Units With No Auto Available Per Total Number of Housing Units Available in Each Census Tract Number of Families Below variable Name Independent variables Persons Over 65 Years Old 10 Table 1 -- Continued variable Code Name Oldfolks Description the Census Poverty Level Per Family in Each Census Tract Bersons Over 65 Years Old Per Capita in Each Census Tract 11 combining the number of families with only one adult as head and divid- ing by the total number of households in each tract with heads. Thus, a rate is arrived at, but not a per capita one. The "No Available Auto" variable is also not a per capita variable. It is calculated by divid- ing the number of housing units with no auto available to them by the total number of housing units in each tract. 1970 Census of the Population data was used for all variables ex- cept three. These are: "Runs" (ambulance runs), "Cops" (police calls for service), and "Crash" (actual number of auto accidents). The "Runs" variable data was collected directly from the Lansing Fire Department records. Their computer requires data to be tabulated by census tract, so it was a very simple matter to acquire the runs per census tract for 1975 and then calculate the variable using 1970 census tract population figures. "Cops" was a bit more difficult to calculate and the police department was very accommodating in helping to provide information. First, since the police keep records by reporting district, it was nec- essary to get records of all police calls for service from each report- ing district. This data was not readily available and a special re- quest for this data for 1975 was made. When the results returned, they were listed by police reporting district. Next, since reporting dis- tricts are smaller than the census tracts, they had to be aggregated in- to groups which approximated the census tracts. In most cases, report- ing districts could be grouped so that several together would exactly make up a census tract. In the instances where some reporting districts had to be split, they were divided into thirds or quarters by visually estimating from a map comparing reporting districts to census tracts. 12 Thus tract 1 might be comprised of reporting areas 1,2,3, and 1/3 of 4. Police calls for service were assigned to tracts in the same manner with tract 1 getting all the calls for service from reporting dis- tricts 1.2.3, and 1/3 of the calls from 4. In this way, police calls for service were assigned to all the tracts and the variable was then calculated using 1970 census figures. The “Crash" variable was obtain- ed by using data on where all accidents occurred in the city in 1975. This data was gathered from the City of Lansing, Traffic Engineering Department. Each location on this list was then plotted on a census tract map and the totals for each tract were then tallied up and record- ed. The actual number of accidents by census tract make up the "Crash" variable. The mixing of 1970 and 1975 data is discussed in the Prob- lems section of the Results chapter. The independent variables were chosen because they were suggested as being important either by employees of Lansing Fire Department and/ or Lansing Mercy Ambulance Service or were suggested in the Aldrich, et a1., study (see footnotes 26 and 27). These variables are: "Fe- males," "Persons Under 10 Years Old," "Females Over 65 Years Old," "Mel es Over 65 Years Old," "Non-Whites," "Divorced Females," "Divorced Males," "Persons Over 65 Years Old," "One Adult Headed Families," "Housing,9 "Auto Accidents,” and "Unemployment". "Females Over 65 Years Old," "Males Over 65 Years Old," and "Persons Over 65 Years Old" may seem like the same variables, but were all included to see if both males and females over 65 influenced ambulance runs or whether it was just the males or just females of this group who were the most influential. In a preliminary study which the author did on the Lansing-East 11 Lansing ambulance system, it was felt that, given the seeming preva- lence of ambulance runs to tracts which represented lower socio-econo- mic class, it might be of interest to include median income and median school years completed variables. These variables also tended to re- late to the ambulance run data and, therefore, were included in this study. The "Poverty" variable was included as another type of economic variable. Median income, if it relates to ambulance runs, can show a trend toward higher ambulance use with an increase or decrease in me- dian income, but the "Poverty" variable can indicate whether or not there may be an increase or decrease in ambulance runs at a certain in- come level. The 1969 poverty level, used in the 1970 Census, ranged from $6,116 for a non-farm family with a male head and seven persons to $1,487 for a single female, 65 years old and over, living on a farm. The poverty level for a non-farm family of four with a male head was $3,745. The "One Adult Headed Families" variable and the "No Available Au- to" variable were-included in this study because these variables may be helpful in explaining the occurrence and location of ambulance runs. The Aldrich, et a1., study indicates that single women (including un- married, separated, and divorced) usually make less use of ambulances than do unmarried, separated or divorced males. The "One Adult Headed Families" variable was included to see if a family might have some in- fluence on the tendency of single adults, especially males, to use the ambulance more. The "No Available Auto" variable was included since a predominance of households in a census tract without an auto available to them might 14 be helpful in explaining why many ambulance runs go to certain tracts. Because these households would have no other means of transportation to a hospital if a medical emergency arose, they might be more inclined to call an ambulance. Professor Keith Honey, School of Urban Planning and Landscape Ar- chitecture, Michigan State University, suggested that a crime indicator be included in the study to see if crime may influence ambulance runs. Therefore, the "Police Calls for Service" variable was included in the study. memes: Data was collected only for the tracts which fire department ambu- lance run data coincided with. Ambulances respond to calls from Lansing tracts 1-30, 31.02, 32, 33.01, 33.02, 34, 36.01, 36.02, 37, 38.01, 44.01, 51, 52, 53.01, 53.02, 55, 202, 214, and 238. Insufficient census data existed for tracts 30, 55, 214, and 238, because of very low population base or the segment of these tracts which fall within the city limits had a very low pepulation base. Therefore, these tracts were elimina- ted by the computer in its calculations when the "pair-wise deletion" command was not specified. This command tells the computer to make the calculations commanded using all cases and to simply skip over the cases of missing data which some tracts had. Therefore, if this command was used, all remaining tracts (44 of them without 30, 55, 214, and 238) would be used in these calculations. When "Pair-wise deletion" is not specified, the computer eliminates all tracts with any missing data. In this case, tracts 14, 35, and 44.01 are eliminated also, since these tracts had at least one variable which could not be calculated due to 35 missing data. Thus, the computer makes its calculations on 41 cases (tracts) in this instance. When calculations are made in this manner, the results are more accurate, since the calculations are made on data which is complete. The reliability of results when the other method is used cannot be guaranteed and it is suggested that "Pair-wise deletion" not always be used.29 The regression equation was calculated using 41 census tracts or about 852 of the tracts which city ambulances respond to (see figure 4). The elimination of three tracts of the 44 (14, 34, and 44.01) eliminates about 61 of the ambulance runs for 1975. The results, then, are based on approximately 942 of the runs generated in 1975. It should be noted here that those tracts which were eliminated by the computer in the computation process (14, 14, and 44.01) had high am- bulance runs per person ratios as well as high ratios for some other variables. This may have had a bearing on the ultimate results, but these tracts were missing data and, to insure the accuracy of the gen- eral equation, had to be eliminated. It is unfortunate that tract 14 was eliminated, especially since that is the downtown area tract and every city has a downtown area which should be included in calculations of this type. However, the tracts which were used in the calculations most likely represent a cross section of what the city of Lansing is generally like. The downtown area most likely has a much larger ambu- lance run rate because few people live there but many people work and 29 Norman H. Nie, C. Hadlai Hull, Jean C. Jenkins, Karin Steinbren- ner, and Dale H. Bent, Statistical Package for the Social Sciences, second edition, version 6.0, (McCraw-Hill, Inc., 1975), p. 353. 36 33.01 32 33.02 3 I Tracts eliminated l S 7 by lack of Census data 4 Not Shown: 238.35 . I. \\\ Tracts eliminated in computing process I. 2' Figure 4 LANSING CENSUS TRAC'I'S NOT INCLUDED IN STUDY 17 shOp there. Another type of model better suited to projecting downtown ambulance run rates must be develOped before the downtown run rate in Lansing can be accurately projected. The Lansing Model developed here is more suitable for making projections for residential areas. Summar This study makes use of census tract information for variables for the multiple regression equation developed here. These variables were chosen by examining related studies and interviewing emergency medical personnel in the area. Once the variables were chosen, information was gathered on them for each census tract in the Lansing area. Only variables about which complete information on all suggested variables could be gathered were used. Multiple regression techniques were then used to analyze the vari- ables and develop a computer model which could be applied to all census tracts and accurately project ambulance calls for any area on a census tract basis. CHAPTER IV RESULTS Introduction This study was carried out with the intent of using census tract data to project ambulance utilization by census tract. The last chap- ter noted the variables which were.included in this study and chapter four discusses these variables, how they influenced this study, and which ones were used in the model and why. A short discussion of the statistical method used by the computer in making calculations is given here to indicate how the correlations and the final regression equation were arrived at. These, as well as some weaknesses of this study, are elaborated on in this chapter. The final section of this chapter is a brief application of the results of the research done in this paper. The cities of Lansing and Grand Rapids, Michigan are used as subjects for application of the re- sults. Computer Method The computer method used is SPSS (Statistical Package for the So- 30 cial Sciences), which was developed at Northwestern University. The SPSS stepwise, multiple regression analysis technique is the method most often used when there are a large number of independent variables. 10 Ibid. 38 19 In stepwise regression analysis, the computer calculates a simple correlation coefficient between the dependent variable and the indepen- dent variables. The computer then selects the independent variable with the highest correlation and, using this variable, computes a re- gression equation. In the next step, the partial correlation coefficient between the dependent variable and the rest of the independent variables is compu- ted. The partial correlation coefficient is the correlation coefficient between two variables when the effects of other variables have been accounted for. The computer then selects the highest of these and cal- culates a multiple regression equation using the two selected indepen- dent variables. The SPSS program continues in this step by step manner until all the independent variables have been entered into the equation or the computer stops entering variables because their ability to explain additional variation is insignificant. This limit is already set in the SPSS package. The results of each step are printed after each step is completed and they are summarized in a summary table at the end of the computer run e Correlgtions By observing the "Simple R" column in Table 2, one can view the correlation coefficients between the dependent variable "Runs" and the independent variables. This correlation coefficient is known as a sim- ple correlation coefficient and represents the correlation between only two variables--the dependent one and the particular independent one. The partial correlation coefficient is calculated by the computer when Step 1 2 10 11 12 13 14 15 16 17 variable Divmen Skool Income Crash Oldmen Unem Divfem Hous Poverty Nocar Oldfem Singfam Kids Oldfolks Minor Cops Female F. value 113.31840 7.45845 8.64200 6.23848 5.60297 2.62389 1.56660 .73115 .98814 1.46830 1.84073 .34316 .58429 .55328 .10591 .11374 .06963 Dependent Variable . . Runs 40 Table 2 Significance O .010 .006 .017 .024 .115 .220 .399 .328 .235 .185 .563 .451 .464 .748 .739 .794 MULTIPLE REGRESSION SUMMARY TABLE R Square .74196 .78597 .82649 .85212 .87253 .88166 .88702 .88955 .89296 .89795 .90404 .90520 .90721 .90915 .9095? .90996 .91023 Simple R .86253 -.65238 -.53045 .60630 .20492 .34903 .59418 .76110 .50732 .68619 .17923 .31782 -.l9308 .20419 .18995 .77217 - e 18189 41 it is ranking the independent variables in order of importance. As mentioned previously, partial correlation coefficients are calculated taking into account the influence of the other independent variables on the one being calculated. This gives a different answer than a simple correlation coefficient and explains why the variables in Table 2 are not listed in order of decreasing Simple R's. The partial correlation coefficient is a more accurate representation of the relationship be- tween variables, but the simple correlation coefficient can be used to get a general idea of how variables relate to each other. Significance also figures in the ranking of the variables by the computer. The significance figure indicates the chance that calcula- tions using a variable could be faulty. The first four variables in Table 2 have the best significance values. The worst of these four is "Crash" which would have a 1.7% chance of giving a faulty answer if it were used in an equation. If one notices which variables have the highest cross correlation with other variables indicated in Table 3, page 42, the ones which represent some manner of social disorganization seem to stand out. In- come relates inversely to all the variables except "Skool". The di- vorced male and female problem relates highly with other problems such as poor housing and high police calls for service. Unemployment re- lates with high ratios of single person headed families. In short, many social problems seem to be inter-related and compound themselves in certain tracts. This is somewhat logical, since the nature and con- sequences of social problems often seem to cause people who have these problems to locate near one another for economic or social reasons. Female Kids Oldfem Oldmen Income . Caps Minor Skool Crash Divfem Divmen Hous Singfam Nocar Poverty Oldfolks Runs -.18189 -.19303 .17923 .20492 - . 53045 .77217 .18995 - . 65238 .34903 .qoeao .59438 .86253 .76110 .31782 .63619 .50732 .20419 42 Table 3 CORRELATION COEFFICIENTS Female -.17713 .52403 .28437 -.02023 -.12423 .04493 .17657 -.O7279 -.05803 .29792 -.O7216 -.21132 .09473 .33917 .16319 .51625 Kids -.66l97 -.5433S .OO726 -.l4532 .14173 -.18806 .17040 -.21093 -.49217 -.40414 .04125 .32366 -.44787 .06582 ' e 63379 Oldfem .85457 -.32018 .15079 -.O7482 -.l3261 -.OOO73 .36497 .63171 .35384 -.01926 .08366 .61178 .14825 .96607 Oldmen -.40376 .21807 -.O64OS -.33965 .16065 .38281 .53818 .34247 .06526 .23311 .52701 .22651 .91706 Income -.66813 -.43649 -.75199 -.sao9s -.43420 -.62150 -.64111 -.68057 -.70464 4.68434 -.77l.29 -.34967 Cops .33474 .50842 .56241 .63642 .57708 .81264 .81968 .38775 .66811 .64287 .20319 Skool Unem Crash Divfem Divmen Hous Singfam Nocar Poverty Oldfolks House Singfam Nocar Poverty Oldfolks 4.1 Table 3 -- Continued -.37278 .37661 .05973 .23739 .22870 .26948 .73593 .38786 .63230 -.01891 Minor .77638 .29613 .82288 .57738 .36623 Dim-mm- -.62783 -.41506 -.39140 -.55938 -.72242 -.72215 -.45992 -.71178 . e20933 Skool .49514 .55530 .68627 .03062 Hous .26155 .31027 .38428 .65215 .57021 .32765 .54705 .07891 Unem .45802 .79510 .13051 .36250 .52309 .41277 .14999 .39182 .29870 .37999 Crash .68868 .63049 Singfam Nocar .77979 .47976 .36256 .87878 .55381 .60821 Divfem .21775 Poverty 44 With this in mind, it is interesting to see that the ambulance run rate correlates negatively with high median income (-.53). These are simple correlations and can be seen in Table 3. The ambulance run rate correlates positively with high police call rates (.77), high incidences of divorced males (.86) and females (.59), more than average housing problems (.76), and a high "No Auto Available to Household" rate (.64). The "Poverty" variable also correlates highly with "Runs" (51). Unemployment correlated with high ambulance runs at about the .35 level while the "Single Person Headed Families" variable correlated at about the .32 level. The characteristics with high correlations could be associated with social disorganization and lower socio-economic class. In effect, what this means to emergency ambulance planners is that census tracts which exhibit evidence of social disorganization or low socio-economic class will be the ones which also require the most ambulance service. Other variables did not correlate highly with the "Runs" variable. The "Female" variable correlated at the -.18 level, "Kids" correlated at the -.19 level, "Oldfem" correlated at .18, "Oldmen“ at..20, and "Oldfolks" correlated at the .20 level. These are not as significant as the previous correlations, but it is important to note that none of them came out with .00 correlation, below the .10 mark or above the -.10 mark. Before the reliability of any variables can be established, their significance figures must be checked. Just a high correlation does not necessarily insure importance. The significance figures of the 17 variables in this study indicate that only the "Divmen", Skool“, "Income" 45 and “Crash" variables can be used if one wishes to remain above the 981 mark for significance. Reasons why particular variables show a higher correlation and are more significant than others cannot be determined without additional investigation. The similarity between the variables with the highest correlation in their relation to social disorganization has been noted, however. Thg_§guation If one looks at the summary tables on the following pages, one can see the results of entering all 17 variables into the equation to try to explain the variation. The computer notes each step and the amount of variation which this model accounts for is noted after the last variable entered (I? I .91023). Thus, looking at the last R? figure in Table 2, one can see that by entering all the B values listed in Table 4, plus the constant, into an equation for each tract, one can account for more than 91% of the variation in runs between census tracts. The 8 value is calculated by the computer and is used to calculate a multiple regression equation of the type Y’- a + B x1 + 1 B x ., B x + .., + ann‘+ error. The Y value is the actual ambulance 2 2 3 runs per pzrson and the 8 values are those which have been calculated to help estimate the actual number of ambulance runs per person. The x values represent the variable values (per person in each census tract) which the 8 values must be multiplied with to obtain the Y estimate. In the case of a four variable model, there will be four B values and four x values for each census tract. Table 4 also lists the standard error. This is a statistical variable Income Crash (Constant) 46 Table 4 SUMARY TABLE I 0? MODEL VARIABLES 1.4789729 -.97766304E-02 .291453083.os .896088068-04 .946342043-01 Std Error B .17593033 .23436607E-02 .890194438-06 .358765883-04 .24600982E-01 F Significance 70.670198 17.401589 10.719328 6.2384848 14.797604 47 measurement of dispersion which, in this case, applies to the 8 value. It indicates that the B value may fluctuate a certain amount, but on the average, the 8 value will be the 8 value stated in the table. It must be noted that if all 17 variables were used in the equation, the significance factor would be quite high. Since the higher the sig- nificance factor becomes, the greater the chance of error, some of the variables entered at the end of the equation are rather questionable. Because variables representing social disorganization correlate highly with the ambulance run rate, it is likely that these variables could be used to project.the ambulance run rate for census tracts and could then prove of value to emergency ambulance system planners. How- ever, there are many variables which could be chosen to represent social disorganization or low socio-economic class. The author has chosen 17 and these have been listed and discussed in a previous chapter. Be- cause 17 is still quite a cumbersome number to work with when doing calculations using a regression equation, it was decided to limit the equation to the four most significant variables. In addition, the first four variables account for 93.61 of the total variation which the entire equation could account for. Therefore, by adding the last thirteen varia- bles only 6.41 more variation is accounted for. The significance of the first four variables is also such that it makes them reliable enough to base calculations on (the worst one of the top four is "Crash" which has a 98.31 chance of being accurate if used in an equation). It was decided that the first four variables would be used in mak- ing up the model for projecting ambulance demand per person by census trOCt e 48 The equation, then, consists of these four variables: “Divmen" (divorced males over 14 years old per person in each tract in 1970). "Skool" (median school years completed in each census tract in 1970). "Income" (median income in each census tract in 1970). ”Crash" (actual automobile accidents which occurred in a tract during 1975). The B values and their standard errors are listed in Table 4. The significance and R? statistics listed in Table 5 for this equation indi- cate what has been previously stated concerning the reliability and im- portance of these figures. Given the 8 values listed in Table 4, the equation for any census tract would look like this: Let Y - Ambulance calls per person in any given tract. Y I 1.48(Divmen) + (-.Ol)(Skool) + .OOOOO3(Income) + .00009 (Crash) + .09463, the constant. (Numbers are rounded to two decimal places or at the first non- sero‘integer after the decimal.) Substituting the values for census tract 1, Lansing, Michigan, the equation becomes: Y - 1.48(.0138) + (-.01)(ll.3) + .000003(9710) + .00009(43) +.O9463 . .020424 + (-.113) + .02913 + .00387 + .09463 - .035054 (without rounding error, the actual number is .03672138) runs per thousand persons in tract 1 for 1975. since the data is assumed to be based on 1975. The runs per year, or Y estimate for the year, is: 49 Table 5 SUMMARY TABLE II OF MODEL VARIABLES Step variable F Value Significance Multiple R R Square Simple R 1 Divmen 113.31840 0 .86253 .74396 .86253 2 Skool 7.45845 .010 .88655 .78597 -.65238 3 Income 8.64200 .006 .90912 .82649 -.53045 4 Crash 6.23848 .017 .92310 .85212 .60630 Dependent Variable . . Runs 50 .035054 x 2763. the actual population of tract 1 x 96.85 runs for 1975. The actual number of runs for 1975 was 96. When this figure is corrected for rounding error, the estimated runs, or Y estimate, actually is 101.46. Some of the previous figures give different results than the computer does because they are rounded here for the sake of illustration. The actual Y estimates for all tracts considered in this study are given in Figure 5. For complete results for all tracts, see Appendix B. Fig- ure 5 was drawn by the computer to show how far off the Y estimates were from the actual runs per person figure for each tract. See the list of conversions from observation to census tract in Table 6 to determine which observation refers to which tract. The four variables in the final regression equation all have rather high simple correlation coefficients between them, the highest being between the"Skool" and "Income“ variables (.75188). The lowest correlation between any of the top four variables is -.41506, which is the coefficient between the“BSkool" and "Crash” variables. This is an inverse relationship, but the absolute value is what matters. The remainder of the correlation coefficients are somewhere between .75 and .41. The high correlation among the top four variables indicates that they are not completely independent of each other. High correla- tion between variables also signifies that they are indicators of social disorganization and/or socio-eeonomic poverty. However, the four varia- bles in the equation are effective in projecting the ambulance run rate by census tract. The correlation between them should be kept in mind when the results are being considered, though. IUIIIIUUOUIUU U C S S I 0 N ' P. G -230 L f I P L E E M U I.¥.¢ I . U C C . I C .OOOOIO 0250 I." e .IDUAL .,-ae I. ‘ - V VALUE Y FSTIHJIE 3 OBSERVATION e-o—eo-eHe-ao—m-aHHO-HHH He-fi-O h'NNNNM‘v-INNNNPJ Nv-KV 2:955:5r5799uc.)or3 90': Il'|||ll'lllllll MIL"ULJLL‘..I.I1'LI'| '1" fut-NU LL‘EAJU' mma-‘N‘J‘KP‘OJ—ONC not" c39mkk'nmcomofi‘ cac W:u£«0=0‘.9cf\ltu.:k ~00 HCJNIQJLDeQNHU‘U‘NM «Na: ~01”; JCNQd-d‘m-FM me‘m ante-«KKK NC (ONMS‘ cs 3‘:- nn-lumms 3 .0! rhino.) (“can dddd—edr‘dH—J—Q—s.‘ Add ""H" az-r"W'_‘)l=")al‘ 65°" II.|I'I||I.CIOII I.'L‘J'|~.u".‘u.'d..'":."_.,U .IAJW~ tacafihf‘wm'I‘Nfichcc-flbsfi urn-«C Hrfimocxxl: (\lhf‘JtOU" .40.».mmcz-um—emtrws (Dd-c MN ~c¢~cmmmaus~c~tb3nrs 5N603'9'Ulnh ”1.9.40.- .? {Hen ,c-uf‘h «4.2m: m. -u mans mg- w-B‘ chJnasaflWm: crinnfl Oe..eeo.eee..pd... (e. V" H 1’ C dddndMo-C—I—s—ea-erfi—Hd—e a .JDO'?L¥§'_’C..‘IJ'..Q&.‘ LJud nllltloolliei';lll “Jonah-M315Luau-Mauxwaawo PW;°€""*J'D" Hoar-awash :JDULV'.‘D.‘L>"D;.:3'J "Dz. oowocacrmaaofizca 9:. L C‘Lfi“‘ 'I H'D'UK" \J' ‘,,,( "‘4. was .oloN-«nutNNG‘hb-lr-QN :rl‘fikanft’Ov-anlftf no“ MONNMFMWNM‘MO‘JOH‘ .‘C...’......u-... ................. WMJIAMBO‘DV‘NDJm-sk *ddv‘H—sd O o O HWHHO-m—‘HHHHHHHHH HHHH Hun—".054 O udNNdNWMNNMN|V”em NN'N’T F‘NN-fl‘v Danna-3.13;...zsc‘oat‘ar-o 9.409 92"1 31-! ottlulllglulllou Igll IOI|I wuu'uluwwmwwwuuiuw (“:ou uJ LULL mum) mc30um.nrcmuoo-4mru 9"“me current? m0403509mocmmkwatc 1:00 «30.90:: amnion“): mommcokmkd‘ rout-J “(Pedro KdeuUde‘NJQNOo-e.’ UMDU‘N LO“ :mr- NJN-«mo-awornmu ”H :n .4me hw'rnn (camel-4r. med—“DUNFJ'Z‘CC 7'" kaat'» wo'ok: vac ,dm 6N6 swamp (Navel-F N‘J‘to 9.1 owe-.4:- O........0.0...O Q... .0... ddd—iu—i—nadd-d—sa—s—od due-owl «odes-4H firmn-mrxsnt‘qmncum:~ -_* rec 0‘13“: illouloooulcotol VIII lll‘l IU<...;uJ'u._“Lug..L-_Iu".l'u(qua-mud.» ouggu u. m|_,_Lj“0|lI‘.,J NRDDROOMV‘NJ.¢B.I®MC'GQLCFJVI< mot-"NF mDCJNNU‘lnN~)U\C‘N‘JrG‘t—f.QL'Nt—Ndk gem, .mr-aoaG-«e.3'~ scan-am: drum-ruse: gram-«x. MAN «meow-eomck Munro-u: J act‘ru-LL: NKJdCWCNNN303dC‘ {-4050 aerate-su- :d‘UnoO‘sL"?-4¢3¢‘ln|\ (on, :c‘ 3:: g-4M_:.7rcs:r._o_, lnsflnHmNPJNNm-d““INTH—AV‘NIM—I'TNN 0.......OOI.....HOOOOH.OOOO w r U! i' H m- : t C C «fie-4.014Hfldddfiqfdddflpdfififlp—OOHe-lfifi JQQa-Jaauo6u~g'zuub" '."J);;)G LDC.:"JC.. Il’IIO".'.|"I.L~"..'l;lllll umunuJu 'sJ'aJtuu'unuullJouuJ:1...u.;un..::-.Ju 59L.» om ”mnaoonouoc 22‘90 c-L'fi’: 015:. Oo'at‘tsc—wu'aangoout. ‘.\r .pc th‘hao canalsoa'aaoo-aooucct‘. nocnaaus‘Dnr: L‘gfi""£u"‘t"‘ fi:_’tfl'_?(,)"’ )|‘